[3.3] allow user-specified schema in read if it's consistent (#3929) (#…

…4125)

backport #3929 to 3.3

<!--
Thanks for sending a pull request!  Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://github.com/delta-io/delta/blob/master/CONTRIBUTING.md
2. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP]
Your PR title ...'.
  3. Be sure to keep the PR description updated to reflect all changes.
  4. Please write your PR title to summarize what this PR proposes.
5. If possible, provide a concise example to reproduce the issue for a
faster review.
6. If applicable, include the corresponding issue number in the PR title
and link it in the body.
-->

#### Which Delta project/connector is this regarding?
<!--
Please add the component selected below to the beginning of the pull
request title
For example: [Spark] Title of my pull request
-->

- [x] Spark
- [ ] Standalone
- [ ] Flink
- [ ] Kernel
- [ ] Other (fill in here)

## Description

<!--
- Describe what this PR changes.
- Describe why we need the change.
 
If this PR resolves an issue be sure to include "Resolves #XXX" to
correctly link and close the issue upon merge.
-->

User-specified schema may come from the catalog if the Delta table is
stored in an external catalog that syncs the table schema with the Delta
log. We should allow it if it's the same as the real Delta table schema.

This is already the case for batch read, see
apache/spark#15046

This PR changes the Delta streaming read to allow it as well.

Note: since Delta uses DS v2 (`TableProvider`) and explicitly claims
that user-specified schema is not supported
(`TableProvider#supportsExternalMetadata` returns false by default), end
users still can't specify schema in `spark.read/readStream.schema`. This
change is only for advanced Spark plugins that can construct logical
plans to triggers Delta v1 source stream scan.

## How was this patch tested?

<!--
If tests were added, say they were added here. Please make sure to test
the changes thoroughly including negative and positive cases if
possible.
If the changes were tested in any way other than unit tests, please
clarify how you tested step by step (ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future).
If the changes were not tested, please explain why.
-->
a new test
## Does this PR introduce _any_ user-facing changes?

<!--
If yes, please clarify the previous behavior and the change this PR
proposes - provide the console output, description and/or an example to
show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change
compared to the released Delta Lake versions or within the unreleased
branches such as master.
If no, write 'No'.
-->
No

spark/src/main/scala/org/apache/spark/sql/delta/sources/DeltaDataSource.scala

@@ -75,9 +75,6 @@ class DeltaDataSource
       schema: Option[StructType],
       providerName: String,
       parameters: Map[String, String]): (String, StructType) = {
-    if (schema.nonEmpty && schema.get.nonEmpty) {
-      throw DeltaErrors.specifySchemaAtReadTimeException
-    }
     val path = parameters.getOrElse("path", {
       throw DeltaErrors.pathNotSpecifiedException
     })
@@ -108,6 +105,11 @@ class DeltaDataSource
         .getOrElse(snapshot.schema)
     }
 
+    if (schema.nonEmpty && schema.get.nonEmpty &&
+      !DataType.equalsIgnoreCompatibleNullability(readSchema, schema.get)) {
+      throw DeltaErrors.specifySchemaAtReadTimeException
+    }
+
     val schemaToUse = DeltaColumnMapping.dropColumnMappingMetadata(
       DeltaTableUtils.removeInternalWriterMetadata(sqlContext.sparkSession, readSchema)
     )

spark/src/test/scala/org/apache/spark/sql/delta/DeltaSourceSuite.scala

@@ -144,7 +144,9 @@ class DeltaSourceSuite extends DeltaSourceSuiteBase
 
   test("disallow user specified schema") {
     withTempDir { inputDir =>
-      new File(inputDir, "_delta_log").mkdir()
+      val deltaLog = DeltaLog.forTable(spark, new Path(inputDir.toURI))
+      withMetadata(deltaLog, StructType.fromDDL("value STRING"))
+
       val e = intercept[AnalysisException] {
         spark.readStream
           .schema(StructType.fromDDL("a INT, b STRING"))
@@ -154,6 +156,32 @@ class DeltaSourceSuite extends DeltaSourceSuiteBase
       for (msg <- Seq("Delta does not support specifying the schema at read time")) {
         assert(e.getMessage.contains(msg))
       }
+
+      val e2 = intercept[Exception] {
+        spark.readStream
+          .schema(StructType.fromDDL("value STRING"))
+          .format("delta")
+          .load(inputDir.getCanonicalPath)
+      }
+      assert(e2.getMessage.contains("does not support user-specified schema"))
+    }
+  }
+
+  test("allow user specified schema if consistent: v1 source") {
+    withTempDir { inputDir =>
+      val deltaLog = DeltaLog.forTable(spark, new Path(inputDir.toURI))
+      withMetadata(deltaLog, StructType.fromDDL("value STRING"))
+
+      import org.apache.spark.sql.execution.datasources.DataSource
+      // User-specified schema is allowed if it's consistent with the actual Delta table schema.
+      // Here we use Spark internal APIs to trigger v1 source code path. That being said, we
+      // are not fixing end-user behavior, but advanced Spark plugins.
+      val v1DataSource = DataSource(
+        spark,
+        userSpecifiedSchema = Some(StructType.fromDDL("value STRING")),
+        className = "delta",
+        options = Map("path" -> inputDir.getCanonicalPath))
+      Dataset.ofRows(spark, StreamingRelation(v1DataSource))
     }
   }